Component characteristic prediction method, component characteristic prediction program, and component characteristic prediction device

Abstract

To provide a component characteristic prediction method, a component characteristic prediction program, and a component characteristic prediction device capable of efficiently grasping a resin material that satisfies the required characteristics of a component.SOLUTION: A component characteristic prediction method causes a computer to predict characteristics of a component produced by using a resin material, and includes an analysis step of predicting characteristics of a component by CAE analysis using resin composition properties output by a learned model generated through learning with manufacturing conditions of a resin composition as explanatory variables and physical properties of the resin composition as objective variables, and shape data of the component.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a component characteristic prediction method, a component characteristic prediction program, and a component characteristic prediction device.

Background Art

[0002] Components made of resin materials are designed in terms of shape and resin materials (composition, manufacturing conditions, etc.) by predicting component characteristics such as mechanical strength. As a technique for predicting component characteristics, CAE (Computer Aided Engineering) analysis using resin physical properties of resin materials constituting components and component shapes is known (see, for example, Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] Here, conventionally, CAE analysis has been performed using physical properties of known resin materials for which resin physical properties have been measured for CAE analysis. However, it has been desired to be able to grasp a resin material more appropriate for required characteristics.

[0005] The present invention has been made in view of the above, and an object thereof is to provide a component characteristic prediction method, a component characteristic prediction program, and a component characteristic prediction device capable of efficiently grasping a resin material that satisfies the required characteristics of a component.

Means for Solving the Problems

[0006] In order to solve the above-described problems and achieve the object, a component characteristic prediction method according to the present invention is a component characteristic prediction method in which a computer predicts characteristics of a component manufactured using a resin material, and a learning model generated by learning with manufacturing conditions of a resin composition as explanatory variables and physical properties of the resin composition as objective variables outputs physical properties of the resin composition, and an analysis step of predicting characteristics of the component by CAE analysis using the shape data of the component.

[0007] Further, in the component characteristic prediction method according to the present invention, in the above invention, in the analysis step, the CAE analysis is executed using virtual resin composition physical properties and known resin composition physical properties.

[0008] Further, in the component characteristic prediction method according to the present invention, in the above invention, in the analysis step, a resin composition that satisfies set characteristic conditions is extracted from the analysis results by the CAE analysis.

[0009] Further, the component characteristic prediction method according to the present invention further includes a learning step of generating the learned model by learning with manufacturing conditions of a resin composition having known physical properties as explanatory variables and physical properties of the resin composition as objective variables in the above invention.

[0010] A component characteristic prediction program according to the present invention is a component characteristic prediction program that causes a computer to predict characteristics of a component manufactured using a resin material, and causes the computer to execute an analysis step of predicting characteristics of the component by CAE analysis using physical properties of a resin composition output by a learned model generated by learning with manufacturing conditions of the resin composition as explanatory variables and physical properties of the resin composition as objective variables and shape data of the component.

[0011] Further, the component characteristic prediction device according to the present invention is a component characteristic prediction device that predicts the characteristics of a component manufactured using a resin material, and includes an analysis unit that predicts the characteristics of the component by CAE analysis using the physical properties of the resin composition output by a learned model generated by learning with the manufacturing conditions of the resin composition as explanatory variables and the physical properties of the resin composition as objective variables, and the shape data of the component.

Advantages of the Invention

[0012] According to the present invention, it is possible to efficiently grasp a resin material that satisfies the required characteristics of a component.

Brief Description of the Drawings

[0013]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Embodiments for Carrying Out the Invention

[0014] Hereinafter, embodiments of the component characteristic prediction method, the component characteristic prediction program, and the component characteristic prediction device according to the present invention will be described in detail with reference to the drawings. Note that the present invention is not limited by this embodiment. In addition, the individual embodiments of the present invention are not independent, and can be appropriately implemented in combination with each other.

[0015] (Embodiment) [System Configuration] FIG. 1 is a diagram showing a schematic configuration of a component characteristic prediction system according to an embodiment of the present invention. The component characteristic prediction system 1 includes a learning device 2 that creates learning data and generates a learned model learned using the created learning data, an analysis device 3 that executes analysis processing for component characteristic prediction using the learned model generated by the learning device 2, a display device 4 that displays information including the analysis result of the analysis device 3, and an input device 5.

[0016] The learning device 2 is electrically connected to the analysis device 3. The learning device 2 selectively extracts learning data, generates and outputs a learned model by learning using the extracted learning data. FIG. 2 is a block diagram showing the configuration of a learning device included in an analysis system according to an embodiment of the present invention. The learning device 2 includes a learning data generation unit 21, a learning unit 22, a control unit 23, and a storage unit 24.

[0017] The learning data generation unit 21 generates learning data using the data stored in the storage unit 24 or the data input from the outside. The learning data generation unit 21 generates a data set having the manufacturing conditions of the resin composition as explanatory variables and the physical properties of the resin composition as objective variables.

[0018] The learning unit 22 performs learning using the learning data to generate a learned model. At this time, the learning unit 22 performs learning using a data set having the manufacturing conditions of the resin composition as explanatory variables and the physical properties of the resin composition as objective variables to generate a learned model.

[0019] In this embodiment, a component manufactured using a resin composition will be described. The manufacturing conditions of the resin composition include the types of constituent materials, the resin composition (mixing ratio of raw materials), and process conditions. Further, the physical properties of the resin composition are either physical properties having specific values such as specific heat, tensile strength, and weld strength, or physical properties including continuous values such as PVT characteristics showing the relationship between pressure, specific volume, and temperature, melt viscosity measured for each temperature and shear rate, and stress-strain curve (SS curve).

[0020] Here, the resin composition is a material mainly composed of resin. There is no particular limitation on the resin constituting the resin composition. For example, styrene resin, fluororesin, polyoxymethylene, polyamide, polyester, polyamideimide, vinyl chloride, olefin resin, polyolefin elastomer, polyether ester elastomer, polyether amide elastomer, polyacrylate, polyphenylene ether, polycarbonate, polyether sulfone, polyetherimide, polyether ketone, polyether ether ketone, polyarylene sulfide, cellulose derivative, liquid crystalline resin, polytetrafluoroethylene resin, epoxy resin, and modified resins thereof can be mentioned. Two or more of these may be contained. Note that the resin composition may contain fibrous fillers, non-fibrous inorganic fillers, and other additives.

[0021] Examples of fibrous fillers include glass fiber, glass milled fiber, glass flat fiber, profiled cross-section glass fiber, glass cut fiber, flat glass fiber, stainless steel fiber, aluminum fiber, brass fiber, rock wool, carbon fibers such as PAN (Polyacrylonitrile)-based and pitch-based carbon fibers, carbon nanotubes, carbon nanofibers, calcium carbonate whiskers, wollastonite whiskers, potassium titanate whiskers, barium titanate whiskers, aluminum borate whiskers, silicon nitride whiskers, aramid fiber, alumina fiber, silicon carbide fiber, asbestos fiber, gypsum fiber, ceramic fiber, zirconia fiber, silica fiber, titanium oxide fiber, silicon carbide fiber, etc. Two or more of these can be used in combination. Among them, glass fiber and carbon fiber are preferred.

[0022] Examples of non-fibrous inorganic fillers include silicates such as talc, wollastonite, zeolite, sericite, mica, kaolin, clay, pyrophyllite, bentonite, asbestos, aluminum silicate, hydrotalcite, silicon oxide, glass powder, magnesium oxide, aluminum oxide (alumina), silica (crushed or spherical), quartz, glass beads, glass flakes, crushed or irregular glass, glass microballoons, molybdenum disulfide, crushed aluminum oxide, translucent alumina (fibrous, plate-like, scaly, granular, irregular, crushed), titanium oxide (crushed), zinc oxide (fibrous, plate-like, scaly, granular, irregular, crushed), etc., calcium carbonate, magnesium carbonate, zinc carbonate, etc., carbonates, calcium sulfate, barium sulfate, etc., sulfates, calcium hydroxide, magnesium hydroxide, aluminum hydroxide, etc., hydroxides, silicon carbide, carbon black and silica, graphite, aluminum nitride, translucent aluminum nitride (fibrous, plate-like, scaly, granular, irregular, crushed), calcium polyphosphate, graphite, metal powder, metal flakes, metal ribbons, metal oxides, etc. Here, specific examples of metal types (metal powder, metal flakes, metal ribbons) include silver, nickel, copper, zinc, aluminum, stainless steel, iron, brass, chromium, tin, etc. In addition, other inorganic fillers include carbon powder, graphite, carbon flakes, scaly carbon, fullerenes, graphene, etc. These may be hollow, and furthermore, it is also possible to use two or more of these inorganic fillers in combination. Among them, calcium carbonate, carbon black, and graphite are preferred.

[0023] Examples of other additives include, for example, silane compounds, antioxidants and heat stabilizers, weathering agents, mold release agents and lubricants, pigments, dyes, crystal nucleating agents, plasticizers, antistatic agents, flame retardants, heat stabilizers, lubricants, ultraviolet light inhibitors, colorants, flame retardants, and foaming agents, etc., ordinary additives.

[0024] The machine learning performed by the learning unit 22 can adopt known machine learning methods. Examples of the statistical models adopted for machine learning include, for example, simple linear regression models, Ridge regression, Lasso regression, Elastic Net regression, general additive models, random forest regression, rulefit regression, gradient boosting trees, extra trees, support vector regression, Gaussian process regression, regression by the k-nearest neighbor method, kernel ridge regression, neural networks, and the like.

[0025] When the learning unit 22 generates a learned model by machine learning using regularization, a plurality of candidate values of the hyperparameters of the regression model are given, and learning is executed for each of the given candidate values of the hyperparameters. Thereafter, for the model learned by each candidate value, using the learning data, the prediction error by cross-validation or hold-out validation is calculated, and the regression model that gives the minimum prediction error is selected. The selected regression model is output as the learned model. Here, the hyperparameters mentioned here are, for example, the number of layers of a neural network and the regularization coefficient.

[0026] The control unit 23 comprehensively controls the operation of the learning device 2.

[0027] The storage unit 24 stores various programs for operating the learning device 2 and data including various parameters necessary for the operation of the learning device 2. The various programs also include a learning data generation program for generating learning data for generating a learned model and a learned model generation program for generating a learned model by learning using the learning data.

[0028] The storage unit 24 is configured using a ROM (Read Only Memory) in which various programs and the like are pre-installed, and a RAM (Random Access Memory), HDD (Hard Disk Drive), SSD (Solid State Drive), etc. for storing the calculation parameters and data of each process.

[0029] Various programs can also be recorded on computer-readable recording media such as HDDs, flash memories, CD-ROMs, DVD-ROMs, and Blu-ray (registered trademark) and widely distributed. The communication network mentioned here is configured using, for example, existing public line networks, LANs (Local Area Networks), WANs (Wide Area Networks), etc., regardless of whether it is wired or wireless.

[0030] The learning device 2 having the above functional configuration is a computer configured using one or more hardware components such as a CPU (Central Processing Unit), GPU (Graphics Processing Unit), ASIC (Application Specific Integrated Circuit), and FPGA (Field Programmable Gate Array). Also, the learning device 2 is connected so as to be able to transmit and receive information to and from the analysis device 3 and the input device 5 via a communication network.

[0031] The analysis device 3 is electrically connected to the learning device 2 and the display device 4. The analysis device 3 executes an analysis process for predicting component characteristics using the learned model created by the learning device 2 and outputs the analysis result. FIG. 3 is a block diagram showing the configuration of the analysis device included in the analysis system according to Embodiment 1 of the present invention. The analysis device 3 includes a calculation unit 31, a control unit 32, and a storage unit 33.

[0032] The calculation unit 31 calculates component characteristics by CAE (Computer Aided Engineering) analysis using the predicted value of the physical properties of the resin composition output by inputting the manufacturing conditions of the resin composition into the learned model and the shape data of the component to be manufactured. Component characteristics include, for example, mechanical strength, in-mold fluidity, warpage, shrinkage, etc. A known method can be adopted for the CAE analysis. Also, when the calculation unit 31 uses an analysis model as the shape data, this analysis model is created using, for example, three-dimensional CAD (Computer Aided Design) data.

[0033] Here, as CAE analysis, there are injection molding analysis, extrusion molding analysis, additive manufacturing analysis for predicting characteristics during molding into the part shape, and structural analysis, structural optimization analysis, thermal management analysis, vibration analysis, electromagnetic field analysis, optical analysis, etc. for predicting the characteristics of the resin in the part shape. Further, after performing the characteristic analysis during molding, an analysis may be performed by combining the predicted results and the characteristic analysis in the part shape.

[0034] The control unit 32 comprehensively controls the operation of the analysis device 3. The control unit 32 has a display control unit 321 that causes the display device 4 to display the calculation result (analysis result) of the calculation unit 31. The display control unit 321 may cause the display device 4 to display information regarding analysis conditions, manufacturing conditions, etc. in addition to the analysis result.

[0035] The storage unit 33 stores various programs for operating the analysis device 3 and data including various parameters necessary for the operation of the analysis device 3. The various programs also include a component characteristic prediction program executed using a learned model. The storage unit 33 is configured using a ROM in which various programs etc. are pre-installed, and a RAM, HDD, SSD, etc. that store calculation parameters and data for each process.

[0036] The various programs can also be recorded on a computer-readable recording medium such as an HDD, flash memory, CD-ROM, DVD-ROM, Blu-ray (registered trademark), etc. and widely distributed. Further, it is also possible for the analysis device 3 to acquire various programs via a communication network. The communication network mentioned here is configured using, for example, an existing public line network, LAN, WAN, etc., and can be either wired or wireless.

[0037] The analysis device 3 having the above functional configuration is a computer configured using one or more hardware such as a CPU, GPU, ASIC, FPGA, etc. Further, the analysis device 3 is connected to be able to transmit and receive information to and from the learning device 2, the display device 4, and the input device 5 via a communication network.

[0038] The display device 4 is a display composed of liquid crystal, organic EL (Electro Luminescence), etc., and is communicably connected to the analysis device 3. The display device 4 acquires and displays display data output from the analysis device 3 under the control of the display control unit 321. Note that the display device 4 may have an audio output function such as a speaker.

[0039] The input device 5 is connected to be able to transmit and receive information to and from the learning device 2 and the analysis device 3 via a communication network. The input device 5 receives input of various information including information such as settings related to the process of predicting component characteristics, and outputs the received information to the learning device 2 and the analysis device 3. The input device 5 is configured using a user interface such as a keyboard, mouse, microphone, touch panel, etc.

[0040] [Component characteristic prediction process] FIG. 4 is a flowchart showing an overview of the component characteristic prediction process performed by the component characteristic prediction device. When there is an instruction input to predict component characteristics in the learning device 2, first, the learning data generation unit 21 generates learning data, and the learning unit 22 performs learning with the manufacturing conditions of the resin composition as explanatory variables and the physical properties of the resin composition as objective variables to generate a learned model (step S11: learning step). Further, the explanatory variables may include, in addition to the resin composition, process conditions when producing the resin composition, etc.

[0041] Thereafter, the analysis device 3 performs a CAE analysis on the characteristics of the component (step S12: analysis step). In this step S12, the calculation unit 31 calculates the component characteristics by CAE analysis using the physical properties of the resin composition and, for example, shape CAD data.

[0042] FIG. 5 is a diagram for explaining the outline of the analysis process in the component characteristic prediction process. The calculation unit 31 inputs the manufacturing conditions of the resin composition (virtual resin composition manufacturing conditions 100) preset as candidates into the learned model 110, and acquires the estimated physical properties of the resin composition (virtual resin composition physical properties 120). At this time, the calculation unit 31 selects, for example, a learned model that outputs the physical properties to be estimated from among a plurality of learned models. Also, for the resin composition (virtual resin composition) set as a candidate, among the physical properties of the resin composition required for CAE calculation, some are not known (some physical properties are unknown), and the manufacturing conditions may use an existing resin composition. In this case, the CAE analysis is performed in a form in which the physical properties of the known resin composition and the estimated physical properties obtained by inputting into the prediction model are mixed. Here, the virtual resin composition is a hypothetical resin composition that does not have actually measured physical property values (actual measurement values), and only the conditions related to manufacturing such as the types of constituent materials, resin composition (mixing ratio of raw materials), and process conditions are defined.

[0043] Thereafter, the calculation unit 31 performs CAE analysis using the estimated physical properties (virtual resin composition physical properties 120) and the shape CAD data of the component to be manufactured (component shape data 130) to calculate the physical properties of the component (virtual component characteristics 140). At this time, in order to use the physical properties of the resin composition estimated by the learned model, the CAE analysis result is the component characteristics predicted based on the estimated physical properties. Here, it is possible to perform CAE analysis including a resin composition with known physical properties (known resin composition physical properties). Thereby, component characteristics can be obtained for formulations with unknown / known physical properties. Note that a resin composition with known physical properties is used, for example, as learning data. Also, in the CAE analysis process, conversion processing may be executed as necessary on the estimated physical properties of the resin composition to convert the numerical values for analysis.

[0044] The control unit 32 outputs the analysis result by the calculation unit 31 (step S13). The control unit 32 causes the display device 4 to display the analysis result. At this time, on the display device 4, for example, an analysis result indicating the physical properties of the component manufactured by the resin composition for which the analysis process has been executed is displayed.

[0045] In the above-described component characteristic prediction process, an example of executing a series of processes from the creation of the learned model to the prediction of physical properties has been described. However, when a learned model has been created in advance, the analysis device 3 reads the learned model from an external server via the storage unit 33, the learning device 2, or the network, and executes only steps S12 and S13. Further, if a learned model has been created in advance and the physical properties of the resin composition have been estimated and stored in an external server or the like, the analysis device 3 may acquire the physical properties of the resin composition via a communication network or the like and execute a CAE analysis.

[0046] In the embodiment described above, the physical properties of a virtual resin composition are estimated using a learned model, and the component characteristics of components composed of each resin composition are calculated by CAE analysis using the estimated physical properties. According to this embodiment, for resin compositions other than known ones, CAE analysis is performed using the physical properties estimated by the learned model to obtain component characteristics, and since the number of candidate resin compositions increases, it is possible to grasp resin compositions that satisfy the required characteristics of the components.

[0047] (Modification example) Next, a modification example of the embodiment of the present invention will be described. Since the component characteristic prediction device according to this modification example is the same as the component characteristic prediction system 1 according to the embodiment, the description thereof will be omitted.

[0048] FIG. 6 is a flowchart showing an outline of the component characteristic prediction process performed by the component characteristic prediction device according to this modification example. When an instruction input for predicting component characteristics is given in the component characteristic prediction system 1, first, the learning device 2 performs learning with the manufacturing conditions of the resin composition as explanatory variables and the physical properties of the resin composition as objective variables to generate a learned model (step S21).

[0049] After that, the analysis device 3 performs CAE analysis on the characteristics of the parts in the same manner as in step S12 (step S22). The calculation unit 31 calculates the part characteristics by CAE analysis using the estimated physical properties obtained from the learned model and the manufacturing conditions of the virtual resin composition, and, for example, the shape CAD data.

[0050] The analysis device 3 executes a filtering process for extracting a resin composition that satisfies the characteristic conditions from among the part characteristics obtained by the CAE analysis (step S23). The calculation unit 31 extracts, for example, part characteristics that satisfy preset characteristic conditions (required characteristics), and acquires the resin composition corresponding to the extracted part characteristics. At this time, the calculation unit 31 can extract a formulation that satisfies the characteristic conditions from a wide range of resin compositions by performing a filtering process on a resin composition with known physical properties and a virtual resin composition whose physical properties are estimated by the learned model.

[0051] The control unit 32 outputs the analysis result by the calculation unit 31 (step S24). The control unit 32 causes the display device 4 to display the analysis result (here, the extraction result). At this time, on the display device 4, for example, for the resin composition for which the analysis process has been executed, the analysis result indicating the resin composition that satisfies the required characteristics regarding the physical properties of the parts produced by the resin composition is displayed.

[0052] In the modification described above, as in the embodiment, the physical properties of the resin composition are estimated using the learned model, and the part characteristics of each resin composition are calculated by CAE analysis using the estimated physical properties of the resin composition. According to this modification, for resin compositions other than known ones, CAE analysis is performed using the physical properties of the resin composition estimated by the learned model to obtain part characteristics, and since the number of candidate resin compositions increases, it is possible to grasp the resin composition that satisfies the required characteristics of the parts.

[0053] Also, according to the modification, since a filtering process for extracting a resin composition that satisfies the required characteristics is executed, it becomes possible to more easily select an appropriate resin composition.

[0054] (Other Embodiments) So far, the embodiments for carrying out the present invention have been described. However, the present invention should not be limited only by the above-described embodiments. In the above-described embodiments, an example in which the component characteristic prediction system 1 is provided as a separate device from a learning device having a learning function and an analysis device having an analysis function has been described. However, the learning unit and the analysis unit may be provided in the same device.

Description of Reference Numerals

[0055] 1 Component characteristic prediction system 2 Learning device 3 Analysis device 4 Display device 5 Input device 21 Learning data generation unit 22 Learning unit 23, 32 Control unit 24, 33 Storage unit 31 Calculation unit 321 Display control unit

Claims

1. A part characteristic prediction method in which a computer predicts characteristics of a part manufactured using a resin material, comprising the steps of: an analysis step of predicting characteristics of the part by CAE analysis using the resin composition properties output from a trained model generated by training using the production conditions of the resin composition as explanatory variables and the physical properties of the resin composition as objective variables, and shape data of the part; The part characteristic prediction method includes:

2. The analyzing step includes: The CAE analysis is performed using the virtual resin composition properties and the known resin composition properties. The part characteristic prediction method according to claim 1 .

3. The analyzing step includes: Extracting manufacturing conditions that satisfy the set characteristic conditions from the analysis results of the CAE analysis; The part characteristic prediction method according to claim 1 or 2.

4. A learning step of generating the trained model by learning using production conditions of a resin composition with known physical properties as explanatory variables and physical properties of the resin composition as objective variables; The method of claim 1 further comprising:

5. A part characteristic prediction program for causing a computer to predict characteristics of a part manufactured using a resin material, comprising: an analysis step of predicting characteristics of the part by CAE analysis using the resin composition properties output from a trained model generated by training using the production conditions of the resin composition as explanatory variables and the physical properties of the resin composition as objective variables, and shape data of the part; A part characteristic prediction program that causes the computer to execute the above.

6. A part characteristic prediction device for predicting characteristics of a part manufactured using a resin material, comprising: an analysis unit that predicts characteristics of the part by CAE analysis using the resin composition properties output from a trained model generated by training using production conditions of the resin composition as explanatory variables and the physical properties of the resin composition as objective variables, and shape data of the part; A part characteristic prediction device comprising:

Images